Self-organized Chain Formation of Nano-Drones in an Open Space

We propose a method for the chain formation of multiple agents in an open space. Chaining can be considered as a building block for several application scenarios, including exploration, maintaining connectivity, or path formation. The proposed method was designed for a very sensing and computationally constrained robot platform, more specifically for nano-drones as they offer advantages in applications in tight spaces or in the proximity of people. To enable portability to a real platform, the method relies on a range and bearing sensing model with a limited field of view that is susceptible to occlusions, which was implemented both in simulation as well as on the real robot through a camera coupled with LEDs. We analyze the method in the simulation-based study. We show that the method works even in presence of noise in sensing and actuation, which rather than being harmful to the chaining performance has a positive effect. We analyze the performance in terms of quality of final chain formation, and speed of convergence, and how these two are affected by increasing swarm size. Finally, we present its practical feasibility in a robotic proof-of-concept featuring nano-drones.